1. Artificial Intelligence Professor: Liqing Zhang Contact Information: Email: zhang-lq@cs.sjtu.edu.cnzhang-lq@cs.sjtu.edu.cn Tel: 6293 2406

2. Introduction Chapter 1

3. 3 1.1 What is AI? (1) Artificial Intelligence (AI) – Intelligent behavior in artifacts – “Design computer programs to make computers smarter” – “Study of how to make computers do things at which, at the moment, people are better” Intelligent behavior – Perception, reasoning, learning, communicating, acting in complex environments Long term goals of AI – Develop machines that do things as well as humans can or possibly even better – Understand behaviors

4. 4 1.1 What Is AI? (2) Can machines think? – Depend on the definitions of “machine”, “think”, “can” “Can” – Can machines think now or someday? – Might machines be able to think theoretically or actually? “Machine” – E6 Bacteriophage: Machine made of proteins – Searle’s belief What we are made of is fundamental to our intelligence Thinking can occur only in very special machines – living ones made of proteins

5. 5 Figure 1.1 Schematic Illustration of E6 Bacteriophage 1.1 What Is AI? (3)

6. 6 1.1 What Is AI? (4) “Think” – Turing test: Decide whether a machine is intelligent or not Interrogator (C): determine man/woman A: try and cause C to make the wrong identification B: help the interrogator Room1 Man (A), Woman (B) Room2 Interrogator (C) teletype

7. 7 1.2 Approaches to AI (1) Two main approaches: symbolic vs. subsymbolic 1. Symbolic – Classical AI (“Good-Old-Fashioned AI” or GOFAI) – Physical symbol system hypothesis – Logical, top-down, designed behavior, knowledge-intensive 2. Subsymbolic – Modern AI, neural networks, evolutionary machines – Intelligent behavior is the result of subsymbolic processing – Biological, bottom-up, emergent behavior, learning-based Brain vs. Computer – Brain: parallel processing, fuzzy logic – Computer: serial processing, binary logic

8. 8 1.2 Approaches to AI (2) Symbolic processing approaches – Physical symbol system hypothesis [Newell & Simon] “A physical symbol system has the necessary and sufficient means for general intelligence action” Physical symbol system: A machine (digital computer) that can manipulate symbolic data, rearrange lists of symbols, replace some symbols, and so on. – Logical operations: McCarthy’s “advice-taker” Represent “knowledge” about a problem domain by declarative sentences based on sentences in first-order logic Logical reasoning to deduce consequences of knowledge applied to declarative knowledge bases

9. 9 1.2 Approaches to AI (3) – Top-down design method Knowledge level – Top level – The knowledge needed by the machine is specified Symbol level – Represent knowledge in symbolic structures (lists) – Specify operations on the structures Implementation level – Actually implement symbol-processing operations

10. 10 1.2 Approaches to AI (4) Subsymbolic processing approaches – Bottom-up style The concept of signal is appropriate at the lowest level – Animat approach Human intelligence evolved only after a billion or more years of life on earth Many of the same evolutionary steps need to make intelligence machines – Symbol grounding Agent’s behaviors interact with the environment to produce complex behavior – Emergent behavior Functionality of an agent: emergent property of the intensive interaction of the system with its dynamic environment

11. 11 1.2 Approaches to AI (5) – Well-known examples of machines coming from the subsymbolic school Neural networks – Inspired by biological models – Ability to learn Evolution systems – Crossover, mutation, fitness Situated automata – Intermediate between the top-down and bottom-up approaches

12. 12 Computer Sci. and Brain Sci. Information Processing in Digital Computer Computing based on Logic CPU and Storage: Separated Data Processing & Storage: Simple Intelligent Information Processing: Complicated and Slow Cognition capability: Weak Information Process Mode: Logic – Information – Statistics Information Processing in the Brain Computing based on Statistics CPU and Storage: Unified Data Processing & Storage: Unknown Intelligent Information Processing: Simple and Fast Cognition capability: Strong Information Process Mode: Statistics － concepts － logic

13. 13 1.3 Brief History of AI (1) Symbolic AI 1943: Production rules 1956: “Artificial Intelligence” 1958: LISP AI language 1965: Resolution theorem proving 1970: PROLOG language 1971: STRIPS planner 1973: MYCIN expert system 1982-92: Fifth generation computer systems project 1986: Society of mind 1994: Intelligent agents Biological AI 1943: McCulloch-Pitt’s neurons 1959: Perceptron 1965: Cybernetics 1966: Simulated evolution 1966: Self-reproducing automata 1975: Genetic algorithm 1982: Neural networks 1986: Connectionism 1987: Artificial life 1992: Genetic programming 1994: DNA computing

14. 14 1.3 Brief History of AI (2) 1940~1950 – Programs that perform elementary reasoning tasks – Alan Turing: First modern article dealing with the possibility of mechanizing human-style intelligence – McCulloch and Pitts: Show that it is possible to compute any computable function by networks of artificial neurons. 1956 – Coined the name “Artificial Intelligence” – Frege: Predicate calculus = Begriffsschrift = “concept writing” – McCarthy: Predicate calculus: language for representing and using knowledge in a system called “advice taker” – Perceptron for learning and for pattern recognition [Rosenblatt]

15. 15 1.3 Brief History of AI (3) 1960~1970 – Problem representations, search techniques, and general heuristics – Simple puzzle solving, game playing, and information retrieval – Chess, Checkers, Theorem proving in plane geometry – GPS (General Problem Solver)

16. 16 1.3 Brief History of AI (4) Late 1970 ~ early 1980 – Development of more capable programs that contained the knowledge required to mimic expert human performance – Methods of representing problem-specific knowledge – DENDRAL Input: chemical formula, mass spectrogram analyses Output: predicting the structure of organic molecules – Expert Systems Medical diagnoses

17. 17 1.3 Brief History of AI (5) DEEP BLUE (1997/5/11) – Chess game playing program Human Intelligence – Ability to perceive/analyze a visual scene Roberts – Ability to understand and generate language Winograd: Natural language understanding system LUNAR system: answer spoken English questions about rock samples collected from the moon

18. 18 1.3 Brief History of AI (6) Neural Networks – Late 1950s: Rosenblatt – 1980s: important class of nonlinear modeling tools AI research – Neural networks + animat approach: problems of connecting symbolic processes to the sensors and efforts of robots in physical environments Robots and Softbots (Agents)

19. 19 1.4 Plan of the Book (I) Agent in grid-space world Grid-space world – 3-dimensional space demarcated by a 2-dimensional grid of cells “floor” Reactive agents – Sense their worlds and act in them – Ability to remember properties and to store internal models of the world – Actions of reactive agents: f(current and past states of their worlds)

20. 20 Figure 1.2 Grid-Space World

21. 21 1.4 Plan of the Book (II) Model or Representation – Symbolic structures and set of computations on the structures – Iconic model Involve data structures, computations Iconic chess model: complete Feature based model – Use declarative descriptions of the environment – Incomplete – Neural Networks

22. 22 1.4 Plan of the Book (III) Agents can make plans – Have the ability to anticipate the effects of their actions – Take actions that are expected to lead toward their goals Agents are able to reason – Can deduce properties of their worlds Agents co-exist with other agents – Communication is an important action

23. 23 1.4 Plan of the Book (IV) Autonomy – Learning is an important part of autonomy – Extent of autonomy Extent that system’s behavior is determined by its immediate inputs and past experience, rather than by its designer’s. – Truly autonomous system Should be able to operate successfully in any environment, given sufficient time to adapt

24. 24 Intelligent Systems

25. 25 Future Artificial Systems Ubiquitous Computing

26. 26 Text Book: N. Nilsson, Artificial Intelligence: A new synthesis Morgan Kaufman,1998 -- Reference Book: Artificial Intelligence: Structures and Strategies for Complex Problem Solving, 4E 【原出版社】 Pearson Education 【作者】 （美） George F.Luger